5 Measurement Techniques for Atmospheric Trace Gas …scarn/teaching/GE4250/PlattStutz... ·...

5

Measurement Techniques for AtmosphericTrace Gas Concentrations and OtherParameters

Our knowledge about the physical and chemical processes in the atmosphererelies to a large extent on measurements of relevant parameters like tempera-ture or ozone mixing ratio throughout the atmosphere. For instance, the ‘me-teorological parameters’ such as temperature, humidity, pressure, radiationintensities, or wind speed and direction have routinely been measured in theatmosphere. In this book, we focus on measurements of trace gas concentra-tions and quantities related to chemical processes in the atmosphere, includingthe intensity of the radiation field. These are the experimental prerequisitesfor understanding the physico-chemical processes in the earth’s atmosphere.In addition, measurements are made in model atmospheres, e.g. smog cham-bers or environmental chambers. As the name ‘trace constituents’ suggests,the species to be investigated here are present at minute quantities only (seeChap. 2). Therefore, the determination of trace gas concentrations in the at-mosphere constitutes a challenge for the analytical techniques employed. Insummary, measurements of atmospheric parameters are done for a variety ofpurposes:

• Weather forecast and climatological studies• Meteorological research• Climate studies• Pollution control• Monitoring of the state of atmosphere• Atmospheric chemistry research

In recent years, the term ‘chemical weather’ was coined to encompassinformation on the trace gas composition of, say, air pollutants of theatmosphere.

U. Platt and J. Stutz, Measurement Techniques for Atmospheric Trace Gas Concentrations

and Other Parameters. In: U. Platt and J. Stutz, Differential Optical Absorption Spectroscopy,

Physics of Earth and Space Environments, pp. 113–134 (2008)

DOI 10.1007/978-3-540-75776-4 5 c© Springer-Verlag Berlin Heidelberg 2008

114 5 Measurement Techniques for Atmospheric Trace Gas Concentrations

5.1 History of Measurement Techniques

After the discovery of a strong, characteristic smell near electric discharges byvan Marum in 1785 and the identification and naming of ozone by ChristianFriedrich Schonbein in 1839, this species was one of the first atmospheric tracegases to be investigated. An earlier technique relied on the use of ‘SchonbeinPaper’ (Schonbein, 1840), a paper impregnated with iodide (e.g. KI) andstarch. Upon exposure to the ambient air and hence to the ozone dispersed init, the following reactions produced a blue colour:

2H+ + 2 I + O3 → I2 + O2 + H2O (R5.1)I2 + starch → blue colour (R5.2)

Together with a suitable colour scale, semi-quantitative measurements couldbe made. Later, it turned out that the amount of blue colour strongly dependson relative humidity and wind speed, and so these relatively large data setfrom the 19th century could not be used for quantitative studies.

These problems were avoided by another technique, which was also in-troduced during the 19th century. Here, air is bubbled through an arsenite-containing solution, which is oxidised to arsenate (Volz and Kley, 1988):

O3 + AsO3−3

I−/I2−−−−→ O2 + AsO3−4 . (R5.3)

Measurements near Paris during 19th century (1876–1910) [the so-calledMontsouris series (Albert-Levy, 1886)] allowed Volz and Kley (1988) to re-construct the much lower pre-industrial ozone content of the troposphere.

Besides the ‘wet chemical’ techniques, optical absorption spectroscopy wasvery closely linked to early research on atmospheric composition. For instance,Alfred Cornu in 1878 (Cornu, 1879) and Sir Walter Noel Hartley (Hartley,1880) showed that a sharp drop in solar intensity below 300 nm was due toatmospheric ozone. This was later used by several researchers to determinethe atmospheric O3 column density (Fabry, 1913; Dobson and Harrison, 1926(see also Chap. 6); Roscoe and Clemitshaw, 1997).

5.2 The Role of Measurements in AtmosphericChemistry

As mentioned earlier, measurements of atmospheric parameters can serve thegoal of coming to a better understanding of our atmospheric environment;however, this can be done in quite a number of different ways. They can becharacterised by the time scale of measurements, i.e. there may be specialised‘measurement campaigns’ aimed at investigating certain phenomena or ‘long-term studies’ to follow the evolution of chemical changes. In addition, wemight look at the spatial scale of investigation, there may be interest in local,

5.2 The Role of Measurements in Atmospheric Chemistry 115

regional or global scale phenomena. Apart from measurements made primarilyfor scientific purposes, a large number of measurements of atmospheric tracegases are made every day to document the pollution level of our atmosphereand to alert the public of extreme pollution events. In trying to group themeasurement efforts, we could use the following criteria:

5.2.1 Long-term Observations

There is great interest in changes in the composition of our atmospheric en-vironment. Some well-known examples include the rise of CO2 mixing ra-tio directly observed since the 1950s (e.g. Keeling et al., 1970). Since then,many other trace gases were followed, e.g. chlorofluorocarbon species (CFMs),methane, or the thickness of the stratospheric ozone layer (i.e. the total at-mospheric O3 column density).

Therefore, long-term observations are frequently aimed at monitoringgradual changes in the trace gas composition of atmosphere. They includeobservations of:

• trends in stratospheric ozone• change of stratospheric chemistry, e.g. as realised in the Network for the

Detection of Stratospheric Change, (NDSC)• evolution of species supplying halogens to the stratosphere (e.g. CFC and

HCFC species)• trend of the tropospheric ozone mixing ratio (Global Atmospheric Watch,

GAW)• trends of greenhouse gases (CO2, CH4, or N2O)• trends of gases indicating the atmospheric oxidation capacity (i.e. the abil-

ity of the atmosphere to remove trace gases). For instance, ozone, CH3CCl3and 14CO are monitored for this purpose.

In this context, the ‘operator dilemma’ should be noted. The measurementof a given set of species with the same technique over an extended period oftime is frequently not considered a scientific challenge; on the other hand, thesuccess of the data series hinges on the very careful conduction of measure-ments. Here, the psychological side of the project may be as important as thetechnological aspects.

5.2.2 Regional and Episodic Studies

Regional and episodic studies seek to investigate the causes, extent, and con-sequences of air pollution. While routine monitoring is an issue, many funda-mental questions can be investigated only by observations made on a regionalscale. Typical measurement tasks in this context are:


• Monitoring of air pollutants (like O3, SO2, NO, NO2, hydrocarbons)• Investigation of urban plume evolution (e.g. with respect to O3 formation

downwind of source regions)• Study of continental plumes• Observation of the Antarctic Stratospheric Ozone Hole• Investigation of polar boundary-layer ozone loss events (the ‘tropospheric

ozone hole’) (Platt and Lehrer, 1995)

5.2.3 Investigation of Fast in-situ (Photo)Chemistry

Studies in ‘Smog’-Chambers (frequently called ‘Reaction Chambers’ or ‘Photo-Reactors’) allow the suppression of transport processes; thus, the effect ofchemistry alone can be investigated. In fact, the phenomenon of troposphericozone formation was observed in smog-chambers long before the chemicalmechanism (see Chap. 2) was discovered. The main disadvantage of smogchambers is the presence of surfaces; thus, care should be taken to avoid arte-facts that may arise from chemical processes at the chamber walls. In order tominimise these problems, very large smog chambers with volumes exceeding100m3 were recently built; these facilities offer surface/volume ratios below1m−1. However, investigation of fast (time scale of seconds) in-situ chemicaland photochemical processes (see Sect. 2.5) in the open atmosphere allowsus to largely neglect the effect of transport, since transport takes place onlyat longer time scales. Thus, it is possible to study the chemical processes di-rectly in the atmosphere. In particular, this is true for free radical (OH/HO2)photochemistry, where the lifetime of the reactive species is on the order ofseconds.

5.3 Requirements for Measurement Techniques

Useful measurement techniques for atmospheric trace species should fulfilltwo main requirements. First, they must be sufficiently sensitive to detect thespecies under consideration at their ambient concentration levels. This canbe a very demanding criterion; for instance, species present at mixing ratiosranging from as low as 0.1 ppt (mixing ratio of 10−13, equivalent to about2 × 106 molecules/cm3) to several parts per billion (ppb) (1 ppb correspondsto a mixing ratio of 10−9) can have a significant influence on the chemicalprocesses in the atmosphere (Logan et al., 1981; Perner et al., 1987a). Thus,detection limits from below 0.1 ppt up to the ppb range are required, depend-ing on the application.

Second, it is equally important for measurement techniques to be specific(or selective), which means that the result of the measurement of a particularspecies must be neither positively nor negatively influenced by any other tracespecies simultaneously present in the probed volume of air. Given the large

5.4 Grouping Measurement Techniques in Categories 117

number of different molecules present at ppt and ppb levels, even in clean air(see Chap. 2), this is also not a trivial condition.

In practice, there are further desirable properties of a measurement tech-nique, including simplicity of design and use of instruments. In addition, the ca-pability of real-time operation (as opposed to taking samples for later analysis)and the possibility of unattended operations are issues. Further considerationsare weight, portability, and dependence of the measurements on ambient condi-tions. To date, no single measurement technique can – even nearly – fulfill all thediverse requirements for trace gas measurements in the atmosphere. Therefore,many different techniques have been developed (e.g. Clemitshaw, 2004).

5.4 Grouping Measurement Techniques in Categories

An important criterion of the available techniques is, therefore, the degree ofspecialisation; for instance, there are instruments for the measurement of asingle species or parameter (like the thermometer that can measure only thetemperature). Simply speaking, we distinguish between specialised techniques(‘box per species techniques’) and universal techniques capable of determiningmany parameters with a single instrument.

Another fundamental property of instruments is their ability to makein-situ or remote sensing measurements. While in-situ measurements comeclose to the ideal to determine trace gas concentrations in a ‘spot’ in spacethat is usually very close to the instrument, remote sensing techniques allowmeasurements from a large distance, perhaps as far as from a satellite in-strument in the earth’s orbit. On the other hand, remote sensing instrumentsusually average the trace gas concentrations over a large volume of air. Remotesensing techniques always rely on the sensing of electro-magnetic radiation,i.e. they are spectroscopic methods.

To date, a large variety of measurement techniques for atmospheric tracegases (and other atmospheric parameters) have been developed. A few exam-ples of instruments belonging to either category include:

• Gas chromatography (GC, universal technique, in-situ)• Optical spectroscopy (universal technique, in-situ, remote sensing)• Mass spectrometry (MS, universal technique, in-situ)• Chemiluminescence (e.g. for the detection of NO or O3)• Chemical amplifiers for the detection of peroxy radicals (e.g. Cantrell et al.,

1984, 1993; Clemitshaw et al., 1997)• Photoacoustic detection (e.g. Sigrist, 1994b)• Electrochemical techniques• Matrix isolation – electron spin resonance (MI-ESR) (e.g. Mihelcic et al.,

1985)• Derivatisation – HPLC or Hantzsch reaction (e.g. for determination of

CH2O) (Fung and Grosjean, 1981; Nash, 1953)


• ‘Bubbler’ combined with wet chemistry, colorimetry (e.g. West and Gaeke,1956; Saltzman, 1954), or ion-chromatography (IC)

In this context, spectroscopic techniques are a powerful variety. Thesetechniques are highly sensitive, very specific, universally usable, give absoluteresults, and have the potential for remote sensing. It is, therefore, not sur-prising that spectroscopic techniques assume a unique role among many other

Table 5.1. Overview of species of relevance to atmospheric chemistry research andmeasurement techniques

Species UV/vis FT-IR TDLS(IR)

GC MS(CIMS)

Fluoresc.Chemolum.

Other

NO O O +NO2 + O + + MI-ESRa

NO3 + MI-ESR, LIFHNO2 + DenuderHNO3 O O DenuderOH + O LIF

HO2/RO2 ? + LIF, Ch. Ab

H2O2 O + +O3 + O O O Electroch.c

HCHO + O + Derivat.d

RCHO Derivat.d

Alkanes +Olefines + OAromatic + + OCO + +DMS +SO2 + +N2O + +CFC’s + +HXe + Wet. Chem.XOe + O +HOXe ?

Symbols denote: well measurable(+), measurable(O), not measurable(empty field).

Abbreviations: UV/vis, UV/visible absorption spectroscopy; FT-IR,Fourier-Transform IR Spectroscopy; TDLS, tunable diode laser spec-troscopy; GC, gas chromatography; MS (CIMS), mass spectrometry(Chemical Ionisation MS).aMatrix isolation – electron-spin-resonance.bChemical amplifier.cElectrochemical cell.dDerivatisation + HPLC.eX = halogen atom (F, Cl, Br, I).

5.5 Experimental Evidence for the Presence of Radicals in the Atmosphere 119

methods that are in use today. In this chapter, we will, therefore, have a closerlook at spectroscopic techniques.

Today, atmospheric chemistry has a comprehensive arsenal of measure-ment techniques at its disposal, Table 5.1 gives an overview of popular tracegas measurement techniques for a series of key species relevant for studiesof atmospheric chemistry. Among a large number of specialised techniques(like the gas-phase chemoluminescence detection of NO), universal techniquesare of great interest due to their relative simplicity – a single (though per-haps relatively complicated by itself) instrument can measure a large rangeof species – these techniques have gained interest.

5.4.1 In-situ Versus Remote Sensing Techniques

Remote sensing techniques allow detection of the properties of an object froma distance. Using these techniques, the trace gas composition in atmospherecan be measured at a point that is remote from the probing instrument. Ex-amples are LIDAR instruments (see Sects. 5.7.3 to 5.7.5) and observations oftrace gas distributions from space. In fact, most DOAS applications constitutea remote detection of trace gases. In contrast to that, in-situ instrumentationmeasures trace gas concentrations (or other parameters) at a particular lo-cation – ideally at a point in space. There are a series of applications wherelocalised measurements are very desirable, e.g. the determination of strongspatial gradients. Since many trace gases have a strong vertical gradient closeto the ground, the observation of these gradients requires measurements thatare localised at least in one (vertical) dimension. It should be noted, how-ever, that in-situ measurements frequently require relatively long integrationtimes, i.e. they average the concentration over a period (tm) of time. As aconsequence, they will average over the distance dm given by:

dm = tm · vw,

where vw denotes the wind speed.Thus, an integration time of 5 min at 2m s−1 wind speed already translates

to a spatial average of dm = 600m.

5.5 Experimental Evidence for the Presence of Radicalsin the Atmosphere

As an example of the combination of measurements by different technolo-gies to solve a difficult problem in atmospheric chemistry, we briefly dis-cuss both direct and indirect evidence for the presence of free radicals in theatmosphere.

Direct detection of free radicals has proven notoriously difficult becauseof the high reactivity and low concentration of these species in the atmo-sphere. For instance, although OH-initiated reaction chains are central to our


understanding of atmospheric chemistry (see Sect. 2.4), direct experimentalverification of this theory took a long time. From the suggestion of the centralrole of OH in the degradation of trace gases and ozone formation (see Chap. 2)in the early 1970s, it took more than a decade before the first reliable directobservations of atmospheric OH became available (see Table 5.2c). Only dur-ing the last decade have tropospheric OH measurements of very high qual-ity been reported (e.g. Dorn et al., 1996; Hofzumahaus et al., 1996; Brauerset al., 1996; Platt et al., 2002; Bloss et al., 2003; Holland et al., 2003; see alsoTable 5.2). Although these data match the model predictions, they do not yetconstitute a data set that covers all the conditions of the atmosphere. On theother hand, there has been a substantial amount of indirect evidence for the

Table 5.2. Evidence for the presence of OH radicals in the troposphere

Evidence Literature (not intended to be com-plete)

Indirect

Photochemical ozone formation e.g. Crutzen, 1974

Pattern of VOC – degradation e.g. Calvert, 1976

NOX – dependence of peroxideformation

Tremmel et al., 1993

Global budgets of CO, 14CO, CH3CCl3 Volz et al., 1981; Brenninkmeijer et al.,1992; Prinn et al., 1987

Semi-direct

Detection by Chemical IonisationMass-Spectrometry

Eisele and Tanner, 1991; Eisele et al.,1994, 1996; Berresheim et al., 2000

Detection by in-situ oxidation of 14CO Campbell et al., 1979; Felton et al.,1990

Direct

Long-Path Differential Optical Absorp-tion Spectroscopy (DOAS) in the UV

Perner et al., 1976, 1987b; Platt et al.,1988; Mount and Eisele, 1992; Mount,1992; Dorn et al., 1996; Brandenburgeret al., 1998

Laser-Induced Fluorescence (LIF) Hard et al., 1992, 1995; Holland et al.,1995, 1998, 2003; Hofzumahaus et al.,1996, 1998; Brauers et al., 1996; Dornet al., 1996; Brune et al., 1995, 1998;Creasey et al., 1997; Hausmann et al.,1997; Carslaw et al., 1999; Georgeet al., 1999; Kanaya et al., 2000, 2001;Heard and Pilling 2003; Hard et al.,2002; Bloss et al., 2003; Holland et al.,2003


presence of OH radicals in the atmosphere. In addition, a number of tech-niques were developed that, while not identifying the OH molecule directly,gave very good and quantitative measure for OH abundance [like CO, 14CO(Volz et al., 1981; Brenninkmeijer et al., 1992), CH3Cl (e.g. Prinn et al. 1987)].In the following section, these methods shall be referred to as semi-direct.Table 5.2 summarises the available evidence for the presence of OH radicalsin the troposphere. Today it can be concluded that, although direct, unequiv-ocal identification of these species in the atmosphere can only be performedin a few cases; there is an enormous amount of indirect evidence of their pres-ence. Thus, the central importance of HOX for atmospheric chemistry can beassumed to be proven beyond reasonable doubt.

In the case of peroxy radicals, the situation is less satisfactory. To date,there are very few direct measurements of HO2 and some other peroxyradicals by matrix isolation – electron spin resonance detection (see Ta-ble 5.3). Infrared spectroscopic detection of HO2 has not yet been achievedin the troposphere. On the other hand, there are techniques for indirect

Table 5.3. Evidence for the presence of HO2/RO2 radicals in the troposphere

Evidence Literature (not intended to becomplete)

IndirectOzone formation Crutzen, 1974Pattern of H2O2 (org. peroxide) forma-tion

Tremmel et al., 1993

Change in the Leighton Ratio (‘Missingoxidant’)

Parrish et al., 1986

Semi-direct

OH–LIF after titration by NO Hard et al., 1995; Brune et al., 1995,1998; George et al., 1999; Creaseyet al., 1997; Kanaya et al., 2000;Hofzumahaus et al., 1996; Hollandet al., 2003; Bloss et al., 2003

Chemical Amplifier (ROX-Box)Measurements

Stedman and Cantrell, 1981; Cantrellet al., 1984, 1993; Hastie et al., 1991;Monks et al., 1996; Clemitshaw et al.,1997; Carslaw et al., 1997; Perner et al.,1999; Mihelcic et al., 2003

Direct

Matrix-IsolationElectron-Spin-Resonance Detection

Mihelcic et al., 1985, 1993, 2003

Tuneable Diode-Laser Spectroscopy(TDLS)

Werle et al., 1991


detection of HO2 available that rely on the conversion to OH via reactionwith NO (Table 5.3). Hydroxyl radicals can then be detected by the well-developed LIF technique. Another widely used ‘semi direct’ technique em-ploys a chemical amplifier (ROX-Box) to convert NO added to the sampledairflow into NO2. Although this basically simple technique appears to havesevere limitations, it has been demonstrated that it can be used to investigatethe tropospheric HOX system, especially in clean air (e.g. Cantrell et al. 1984;Monks et al., 1996).

The first direct detection of NO3 radicals (by DOAS) in the tropospheredates back to 1979 (see Table 5.4), but the idea of considerable influence ofNO3 on tropospheric chemical cycles only slowly gained acceptance. Sincethere appears to be no simple way to derive the global (or regional) averageNO3 concentration, this quantity has to be determined from long-term mea-surement series, which have recently begun (e.g. Heintz et al., 1996; Plane andNien 1991]. Again, there is considerable indirect evidence for the importanceof NO3 radical reactions in the troposphere; these include (1) the lack of sea-sonal dependence of the NOY deposition (e.g. Calvert et al., 1985), suggesting

Table 5.4. Evidence for the presence of NO3 radicals in the troposphere

Evidence Literature (not intended to becomplete)

IndirectSeasonal variation of the NOY de-position

Calvert et al., 1985

Night-time observation of RO2 for-mation

Platt et al., 1990; Mihelcic et al., 1985;Golz et al., 1997

Night-time degradation pattern ofVOCs

Penkett et al., 1993

Semi-directConversion to NO +chemoluminescence

-

DirectDifferential Long-Path AbsorptionSpectroscopy (LP-DOAS)

Noxon, et al., 1980; Noxon, 1983; Plattet al., 1979, 1981, 1984; Plane andNien, 1991; Heintz et al., 1996; Aliwelland Jones, 1998; Allan et al., 1999,2000; Martinez et al., 2000; Geyeret al., 2001, 2002, 2003

Cavity Ringdown Spectroscopy King et al., 2000; Brown et al., 2001–2003

Matrix-IsolationElectron-Spin-Resonance Detection

Mihelcic et al., 1985, 1993, 2003

Laser-Induced Fluorescence (LIF) Wood et al., 2003; Matsumoto et al.,2005


non-photochemical NOX–NOY conversion via NO3–N2O5 during winter; (2)observation of night-time radical formation (e.g. Mihelcic et al., 1993; Golzet al., 2001); (3) degradation patterns of hydrocarbons (Penkett et al., 1993),details are discussed in Sect. 2.6.

Recently, a substantial amount of direct and indirect evidence has beenaccumulated for the presence of halogen atoms and halogen oxide radicals incertain parts of the troposphere (e.g. Wayne et al., 1991; Platt and Honninger,2003; see also Table 5.5). For instance, the degradation pattern of hydrocar-bons, in particular in the Arctic as well as at remote mid-latitude areas, pointsto a noticeable role of Cl and Br atoms in the oxidation capacity of the tro-posphere. In addition, the events of ozone loss on extended areas in the polar

Table 5.5. Evidence for the presence of halogen radicals in the troposphere

Evidence Literature (not intended to be complete)

Indirect‘Hydrocarbon clock’ observations (Cl,Br-atoms),

Solberg et al., 1996; Jobson et al., 1994;Rudolph et al., 1999; Singh et al., 1996;Wingenter et al., 1996; Ramacher et al.,1997, 1999; Rockmann et al., 1999; Plattet al., 2004

Sudden loss of ozone in the (polar)boundary layer

e.g. Barrie et al., 1988; Platt and Lehrer,1996

Semi-directAtomic fluorescence after titration byNO (BrO, ClO)

Toohey et al., 1996

Chemical Amplifier (ROX-Box) mea-surements

Perner et al., 1999

DirectDifferential Long-Path AbsorptionSpectroscopy (LP-DOAS), detectionof BrO, ClO, IO, OIO

Hausmann and Platt, 1994; Unold, 1995;Tuckermann et al., 1997; Platt andLehrer, 1996; Kreher et al., 1997; Hegelset al., 1998; Wagner and Platt, 1998;Richter et al., 1998; Martınez et al., 1999;Alicke et al., 1999; Allan et al., 2000,2001; Wittrock et al., 2000a; Frieß et al.,2001; Wagner et al., 2001; Matveev et al.,2001; Leser et al., 2002; Stutz et al.,2002; Honninger and Platt, 2002, v.Roozendael et al., 2002; Pundt et al.,2002; Bobrowski et al., 2003; Hollwedelet al., 2004; Honninger et al., 2004b,c;Frieß et al. 2004a; Saiz-Lopez, 2004a,b;Bobrowski, 2005; Zingler and Platt, 2005


boundary layer are best explained by BrO- and ClO-catalysed ozone destruc-tion (e.g. Barrie and Platt, 1997), as discussed in Sect. 2.9.

In summary, we have a considerable body of indirect as well as directevidence for the presence of several types of free radicals in the atmosphere.

5.6 Spectroscopic Techniques

To date, no single measurement technique can – even nearly – fulfill all therequirements; therefore, in a particular application the selection of a techniquewill be based on the specific requirements: What species are to be measured?Is simultaneous determination of several species necessary? What is the re-quired accuracy, time resolution, and spatial resolution? Other aspects to beconsidered are logistic requirements like power consumption, mounting of lightsources or retro-reflectors (see below), or accommodation of the instrumenton mobile platforms.

Spectroscopic techniques can be broadly divided, on one hand, into meth-ods relying on absorption of radiation directed to the sample from some source,or, on the other hand, into spectroscopic analysis of radiation emitted bythe sample itself. Further divisions can be made according to the wavelengthrange used and thus the kind of internal state of the sample molecules theradiation interacts with. Figure 5.1 illustrates these different categories bygiving a ‘family tree’ of spectroscopic techniques. In this section we give abrief overview of some spectroscopic techniques used to analyse atmosphericcomposition.

5.6.1 Microwave Spectroscopy

Spectroscopy in the microwave and sub-mm wavelength range can, in prin-ciple, be used in active and passive configurations. To our knowledge, nodetection of atmospheric gases with active microwave instruments (i.e. withinstruments employing their own radiation sources) has been made. However,the measurements of atmospheric parameters (clouds, precipitation, and tur-bulence) with RADAR techniques are common.

Passive measurements of stratospheric species by microwave and sub-mmwave emission are state of the art (e.g. Janssen, 1993). This technique reg-isters the thermal emission due to rotational transitions of the atmosphericmolecules. For instance, the ClO-molecule radiates at νo = 649.448GHz, cor-responding to 0.46 mm wavelength (181/2 → 171/2 transition of 35ClO), butHOCl is also detectable by this technique. The line width is dominated by colli-sional (pressure) broadening, which by far dominates over Doppler-broadening(ΔνD/ν0 ≈ vmolec/c ≈ 10−6), as described in Sect. 3.6. While this strong vari-ation in line width with pressure, and thus altitude of the absorbing molecule,allows the retrieval of vertical profiles from a thorough analysis of the recorded

5.6 Spectroscopic Techniques 125

Emission

Thermal Induced

Chemical LASER

Optical Measurement of Atmospheric Trace Species

AbsorptionPhysical Process

UV/visible(electronic)

IR(vibrat.)

μ-Wave(rotational)Wavelength

molec. physics

Single/doublePath

Multiple foldedPath

Light Patharrangement

None, RetroMirror

Scattering(air, topographic

Radiation transport

ThermalSource

LASERLight source

Non-Dispersive

SpectrometerDiode-Array

SpectralResolution Fourier-

Transform

Fig. 5.1. ‘Family tree’ of spectroscopic measurement techniques for atmospherictrace species. Absorption spectroscopy can be categorised by the type of lightsource – natural (e.g. sunlight) or artificial (e.g. incandescent-, arc lamps, or lasers),the arrangement of the absorption path, the method of obtaining spectral infor-mation (non-dispersive, dispersive techniques, type of spectrometer, etc.), or othercharacteristics of the particular technique

line shape, it also limits the detection sensitivity at an atmospheric pressureexceeding a few mbars. Thus, for ClO, the best sensitivity is reached in the up-per stratosphere, while detection at lower stratospheric or even troposphericaltitudes (i.e. below about 15–25 km) is usually difficult (e.g. Janssen, 1993;Klein et al., 2002; Livesey et al., 2003). The microwave and sub-mm wave-length range is presently not used for measurements in the troposphere; due tothe relatively large pressure broadening of the lines, it would probably requiremeasurements at reduced pressure.

5.6.2 IR Spectroscopy

IR spectroscopy has been in use for several decades, and was initially de-veloped for the detection of atmospheric CO2 by non-dispersive instruments[‘Ultra-Rot AbSorption’ (German for ‘ultra-red absorption’), URAS]. Moremodern instruments are based on Fourier transform (FT) techniques to mea-sure HNO3, CH2O, HCOOH, H2O2, and many other species on km pathlength multiple-reflection cells (e.g. Pitts et al., 1977; Tuazon et al., 1980).The sensitivity is in the low ppb range; thus, those instruments appear to be


best suited for studies of polluted air. The technique has been applied in threemodes of operation:

1. Active operation, where an artificial broadband light source, usually athermal radiator, is used (Pitts et al., 1977; Tuazon et al., 1980; Galleet al., 1994).

2. Active operation, where a tuneable diode laser serves as light source. Dur-ing the last two decades, tunable diode laser spectrometers (TDLS) havebeen developed as field-usable instruments, which were successfully usedto measure HNO3, NO, NO2, CH2O, H2O2 and other species at sub-ppblevels (e.g. Harris et al., 1989; Schiff et al., 1990; see also Sect. 5.7.1).

3. Passive operation using thermal emission from the trace gases under con-sideration. A particular example is the Michelson Interferometer for Pas-sive Atmospheric Sounding (MIPAS) (Fischer, 1993; Fischer and Oelhaff,1996; Notholt et al., 1997).

5.6.3 UV/Visible Absorption Spectroscopy

At ambient temperatures, thermal emission of UV/visible radiation is com-pletely negligible; nevertheless, active and passive arrangements can be used,the latter technique relying on radiation from naturally hot objects, e.g.the sun.

The most common optical arrangement is shown in Fig. 6.1; the lightsource and the receiving system are typically separated by several kilometres.UV/visible absorption spectroscopy can be used in both ‘active’ and ‘passive’modes; in the active mode, an artificial light source (see Chap. 6) providesthe radiation, while in the passive mode natural light sources, usually thesun or stars, act as source of radiation. The strength of these techniqueslies in good specificity, the potential for real-time measurements, and theabsence of wall losses. In particular, the first property makes spectroscopictechniques more suitable for the detection of unstable species like OH radicalsor nitrate radicals (see Chap. 10). Limitations in the systems using a separatedlight source and receiving system are due to logistic requirements (the needfor electric power at two sites several kilometres apart, but in sight of eachother) in the case of unfolded path arrangements; in addition, conditions ofpoor atmospheric visibility (see Sect. 4.7) can make measurements with thistechnique difficult.

LIDAR techniques, on the other hand, combine the absence of wall lossesand the good specificity with somewhat smaller logistic requirements and thecapability to make range-resolved measurements. In contrast to LIDAR, mostof the above systems can only make point or path-averaged measurements.Unfortunately, this advantage of LIDAR is usually obtained at the expense ofreduced sensitivity (see Sect. 5.7.3).

5.7 Selection Criteria for Spectroscopic Techniques 127

5.7 Selection Criteria for Spectroscopic Techniques

The important technical criteria of spectroscopic instruments are the wave-length region used (see Fig. 5.1), the physical principle (i.e. absorption oremission spectroscopy), the arrangement of the light path (path in the openatmosphere or enclosed – frequently folded – path), or the type of light sourceused. The following techniques are presently employed to measure atmospherictrace gases:

• Tunable Diode Laser Spectroscopy (TDLS)• Photo Acoustic Spectroscopy (PAS)• Light Detection And Ranging (LIDAR)• Differential Absorption LIDAR (DIAL)• Laser-Induced Fluorescence (LIF)• Cavity-Ringdown Spectroscopy (CRD)• Mask Correlation Spectroscopy• Differential Optical Absorption Spectroscopy (DOAS)

While, in principle, most techniques (i.e. TDLS, LIDAR, DIAL, LIF, andDOAS) allow light paths (or sensitive volumina, in the case of LIF) in the openatmosphere, some techniques lend themselves more to enclosed light paths.

5.7.1 Tuneable Diode Laser Spectroscopy (TDLS)

The idea of tunable diode laser spectroscopy is to use a narrow-band, frequency-variable laser source to scan a suitable trace gas absorption structure. Thistype of laser can be frequency tuned e.g. by varying the diode current. Inpractice, a sawtooth-like waveform of the laser current results essentially ina linear scan across a certain spectral interval, typically a small fraction ofa wave number. Thus, in principle, no further wavelength selective elementsare required. Early applications relied on lead–salt (Harris et al., 1989; Schiffet al., 1990; Sigrist, 1994) diode lasers operating in the mid-infrared (about3–30 μm), thus covering the fundamental vibrations of many molecules of at-mospheric interest. The disadvantages of lead–salt diode lasers are their lim-ited commercial availability and the requirement for cooling to temperaturesaround (and frequently below) the boiling point of liquid nitrogen. While diodelasers are very compact and low power-consuming devices, the need for liquidnitrogen cooling or closed cycle He-coolers and vacuum thermal insulation,in practice, offsets these advantages. In recent years, further development oftuneable diode lasers yielded devices operating around room temperature inthe near IR (about 1.6–2 μm); here communication laser diodes operating atroom temperatures are available, which allow very compact and lightweightinstruments (e.g. Durry et al., 1999; Gurlit et al., 2005). On the other hand,in the near IR, typically weaker overtone-bands are available. In addition, re-cently, quantum cascade lasers became available, which can generate narrow


band emission even in the mid-IR (e.g. Kosterev and Tittel, 2002; Jimenezet al., 2004).

During the last two decades, TDLS became field-usable instruments, andwere successfully used to measure HNO3, NO, NO2, CH2O, and H2O2 at sub-ppb levels (Harris et al., 1989; Schiff et al., 1990). In the usual arrangementcoupled to a multiple reflection cell, the strength of TDLS lies in the mobil-ity of the instrument, allowing measurements on board ships and airplanes,combined with high sensitivity. The limitations are due to the necessity tooperate under low pressure (in many applications), thus introducing possi-ble losses at the walls of the closed measurement cells. Moreover, at present,diode-laser technology is still quite complex. Therefore, the development oflaser diodes is in demand, and diodes are mainly developed for applicationswhere a mass market is seen (CD players, fibre optical communication, etc.).While the developments in the latter areas are quite impressive, diodes op-timised for spectroscopic purposes, especially in the visible and ultravioletspectral ranges, are still difficult to obtain.

5.7.2 Photo Acoustic Spectroscopy (PAS)

Photo Acoustic Spectroscopy is based on the detection of the pressure changeoccurring as a consequence of the temperature rise due to the energy absorbedby trace gas molecules illuminated by radiation at suitable frequency. Obvi-ously, the absorption of radiation by atmospheric trace gases is very minute,thus giving rise to only small temperature and pressure changes. On the otherhand, extremely sensitive devices for detecting periodic atmospheric pressurechanges are available in the form of microphones for the detection of soundwaves (e.g. Sigrist, 1994).

Thus, a typical PAS instrument consists of an intensity-modulated lightsource – usually a laser – illuminating the interior of a cell (e.g. a tube of afew millimetres of inner diameter and a fraction of a meter length) equippedwith a microphone ‘listening’ to the periodic pressure changes occurring atthe frequency of the modulation. The weak signal is sometimes enhanced bytuning the modulation frequency to an acoustic resonance of the absorptioncell, leading to a signal increase by the ‘quality factor’ Q of the acousticresonator (Sigrist, 1994).

Note that photoacoustic detectors directly measure the difference I0 − I(approximately proportional to D · I0 and thus the trace gas concentration)rather than determining I and I0 separately.

5.7.3 Light Detection And Ranging (LIDAR)

The term LIDAR was coined to resemble the well-known RADAR (RAdiowaveDetection And Ranging) (e.g. Rothe et al., 1974; Hinkley, 1976; Svanberg,1992; Sigrist, 1994). In principle, short pulses of a strong, collimated lightsource, typically a pulsed laser, are emitted into the atmosphere. By analysing


the temporal evolution of the intensity backscattered from the atmosphere,the spatial distribution of scattering and extinction along the direction of theemitted (and received) radiation can be deduced, as illustrated in Fig. 5.2.

The distance from which the radiation intensity (under consideration) isscattered back is given by:

R =c · t2

,

where c is the speed of light, and t is time (after emission of the laser pulse)at which the signal reaches the detector.

The general LIDAR equation is:

E (λ, R, ΔR) = K · E0 · cS(R) σSRΔR

R2·

× exp

⎛⎝−2

R∫

0

[σA (r) · cA (r) + σS (r) · cS (r)] dr

⎞⎠ , (5.1)

with the following meanings to the terms:

E(λ, R, ΔR) denotes radiation energy received from a volume of air betweenR, R + ΔR;

ΔR is the distance interval to be averaged over (corresponds to Δt = 2ΔRc );

K is a constant of the system (describes the influence and size of the receivingsystem);

E0 is the radiation energy emitted by the laser;cs(R)is the concentration of backscattering centres (molecules, aerosol par-

ticles);σSR is the backscatter cross-section;σA is the absorption cross-section (aerosol or gas);cA(R) is the concentration of the absorbers (i.e. gas molecules);σS is the total scattering cross-section (note that σS �= σRS !).

LASER

R

Detector Trace Gas orAerosol - Cloud

Fig. 5.2. The principle of the LIDAR technique


It is worth noting that the returned LIDAR signal is proportional to R−2,while the returned RADAR signal varies with R−4. This is due to the factthat the scattering volume increases with R2, while in the case of RADARthe object (e.g. an airplane) is of constant size, and thus both the transmittedsignal arriving at the target and the returned signal reaching the receiverdecrease proportional to R−2.

The LIDAR techniques combine the absence of wall losses with the uniquecapability to make range-resolved measurements along the line of sight. Incontrast to that, the systems discussed earlier can only make point measure-ments or path-averaged measurements. In the form described up to this point,LIDAR systems have been successfully used to map aerosol distributions.There are, however, two central problems with LIDAR systems:

1. The signal [see (5.1)] depends on backscattering as well as on total scat-tering (or extinction) properties of the atmosphere, which are (in thecase of Mie scattering) difficult to separate (the ratio σS/σSR is usu-ally unknown and must be estimated from a-priori information unlessthe Raman–LIDAR technology is used).

2. The backscattered signal is usually very weak, and thus even when high-energy lasers are used, only a relatively small number of photons is re-ceived. Therefore, the above advantages are usually obtained at the ex-pense of high sensitivity.

5.7.4 Differential Absorption LIDAR (DIAL)

In order to measure the distribution of trace gases, at least two differentwavelengths are required (in contrast to the ‘ordinary’ (= aerosol-)LIDAR)(Svanberg, 1992; Sigrist, 1994). The DIAL equation is obtained by dividingtwo general LIDAR equations (5.1) for two different wavelengths λ1 and λ2,respectively:

E (λ2, R)E (λ1, R)

= exp

⎛⎝−2 (σA (λ2) − σA (λ1)) ·

R∫

0

cA (r) dr

⎞⎠ . (5.2)

Thereby it is assumed that σSR and σS are the same for λ1 and λ2, respectively.This assumption appears justified as long as Δλ = λ2 −λ1 is sufficiently small(a few nanometres). Obviously, the two wavelengths are chosen so that (whiletheir difference is as small as possible) the difference in the absorption cross-sections Δσ = σA(λ2) − σA(λ1) is as large as possible, i.e. one wavelength isat the centre of an absorption line, the other besides the line.

5.7.5 White Light LIDAR

Recently, suggestions and first experiments of two varieties of the LIDARtechnique were made using broadband ‘white light’ sources in conjunction


with wavelength dispersing detectors making use of the DOAS principle, seeSect. 5.7.9. If these techniques prove viable, they could combine the advantagesof LIDAR and DOAS. Presently, there are two approaches:

1. A LIDAR principle that combines the differential UV-visible absorptionspectroscopy and the classical LIDAR technique. The novel element ofthe system is the use of an imaging spectrometer in conjunction witha two-dimensional CCD detector array to simultaneously spectrally andtemporally resolve backscattered radiation. The ‘white’ light (i.e. broad-band radiation) required for this LIDAR approach is provided either by aflash-lamp or a broadband dye laser (of 10–20 nm full width at half maxi-mum tuneable across the UV-visible spectral region). Thus, simultaneousrange-resolved measurement of multiple molecular species is possible bythe DOAS technique. A first application demonstrated absorption stud-ies in the spectral regions where NO3 and H2O absorb using both elasticand inelastic (Raman) backscattered radiation (Povey et al., 1998; Southet al., 1998). In addition, the technique has applicability for a wide range ofmolecules, including O3, NO2, and other spectrally structured absorbers,and for atmospheric temperature sounding, which may be derived from ei-ther rotational Raman return or temperature-dependent absorptions suchas those of O2.

2. In this approach, ultra-short (femtosecond) laser pulses produce a ‘beam’of white light in the atmosphere, which can then be observed with a sepa-rate wavelength-dispersing detector system. Atmospheric constituents willleave their spectroscopic fingerprint on the distance between the ‘beam’and the detector, which can then be detected by DOAS techniques (e.g.Rairoux et al., 2000; Kasparian et al., 2003).

5.7.6 Laser-Induced Fluorescence (LIF)

Laser-Induced Fluorescence is a very sensitive and specific technique for thedetection of atmospheric trace gases. It relies on the excitation of trace gasmolecules by absorption of (laser) radiation, the frequency of which cor-responds to a suitable transition from the molecule’s ground state to an(electronic) excited state, see Figure 3.6. The number of fluorescence photonsdetected is then proportional to the atmospheric trace gas concentration.

A central problem with LIF is the separation of Rayleigh scattered excit-ing radiation from the fluorescence signal. While Rayleigh scattering cross-sections of air molecules (i.e. O2 and N2) are typically six to eight orders ofmagnitude smaller than trace gas absorption cross-sections, typical trace gasmixing ratios are 10−12 to 10−9. Combined with the fact that fluorescenceefficiencies can be orders of magnitude below unity, scattered excitation ra-diation must be suppressed by many orders of magnitude. Several techniqueshave been developed to achieve this goal, which include:


1. Excitation at higher (vibrational) levels above the excited state in whichfluorescence can be observed. The detected wavelengths (red) are thusshifted from the excitation wavelength, and can be separated by suitablefilters (wavelength filtering).

2. Use of short laser pulses for excitation. The fluorescence signal can berecorded after the excitation signal has decayed and thus separated. Thistechnique frequently requires low pressure in order to avoid too rapida decay of the fluorescence signal by collision quenching and hence itsdenomination ‘temporal’ or ‘baric’ filtering (Hard et al., 1979).

Other potential problems in the technique are photochemical formationof species to be detected by excitation radiation [e.g. OH by O3 photolysis(Ortgies et al., 1980; Davis et al., 1981; Shirinzadeh et al., 1987)], or saturationof the transition under consideration. In addition, absolute calibration from‘first principles’ has been found to be extremely difficult (e.g. Hofzumahauset al., 1996, 1998; Holland et al., 2003), and so experimental calibration isnecessary (e.g. Hofzumahaus et al., 1996, 1997, 1998; Schultz et al., 1995).

5.7.7 Cavity-Ringdown (CRDS) and Cavity EnhancedSpectroscopy (CEAS)

Cavity-Ringdown and Cavity Enhanced Spectroscopy are relatively new tech-niques that make use of the effect of atmospheric absorbers on the qualityof a usually passive optical resonator (a ‘cavity’), see also Sect. 7.8.3 (e.g.Paldus and Zare, 1999; Brown, 2003). After initially exciting the resonatorwith a laser and then switching off the laser, the decay (or ‘ringdown’) ofthe resonator is observed, which is influenced by the extinction of light dueto trace gases present in the resonator cavity. Alternatively, if excited withcontinuous radiation, the resonator will exhibit certain attenuation, which canbe monitored. The energy loss per ‘round trip’ of the radiation in an opticalresonator with no absorber is given by the reflectivity of the two mirrors, seeSect. 7.8.3. In the case of very weak absorbers, the quality, Q, of the cavityis only determined by mirror reflectivity, RM, which can reach 0.9999; thus,effective light path lengths on the order of many kilometres can be reachedunder favourable conditions with resonator lengths below 1 m. On the otherhand, in some respects, the limitations of CRDS are similar to the limitationsof DOAS in that the effective light path will be limited by visibility. CRDShas been developed into a useful tool for laboratory measurements; severalmeasurements in the atmosphere, mostly of NO3 radicals, were reported (e.g.King et al., 2000; Ball and Jones, 2003; Brown et al., 2001–2003; Simpson2003; Ball et al., 2004).

5.7.8 Mask Correlation Spectroscopy (COSPEC)

Mass Correlation Spectroscopy was originally conceived as a tool for oilexploration by the detection of atmospheric iodine (I2) vapour, which was


assumed to be frequently associated with oil deposits. The fundamental ideaof COSPEC is to analyse the absorption features (i.e. absorption bands) ofatmospheric constituents imprinted on scattered sunlight. In modern nomen-clature, it is a ‘passive’ instrument (see Chap. 6). The instrument allowedsome degree of remote sensing of atmospheric trace gases. While the originalidea of I2-vapour detection did not prove practical, other applications werefound quickly, which included the observation of tropospheric column densi-ties of SO2 and NO2 in industrial emission (e.g. Giovanelli et al., 1979; Beilkeet al., 1981; Redemann Fischer, 1985) or volcanic plumes (Hoff, 1992).

The COSPEC instrument was originally designed in 1960s (Barringeret al., 1970; Davies, 1970; Davies et al., 1975); it uses an innovative opto-mechanical correlator (e.g. Davies, 1970; Millan and Hoff, 1977; Millan, 1972,1978, 1980) to identify and quantify the spectra of atmospheric trace gases.In this way, a remarkably compact and reliable instrument could be built,since spectral recognition of the molecules to be measured was done by spe-cial opto-mechanics (Fig. 5.3) with relatively simple electronics and withouta computer. Since neither fast-scanning, multi-channel detectors nor powerfulcomputers were available at that time, the mask correlation technique was anattractive way to measure total columns.

By making the instrument mobile and horizontally traversing emissionplumes, a total integrated concentration cross-section of the plumes can beobtained (Giovanelli et al., 1979). After multiplication with the concentra-tion weighted wind component perpendicular to the cross-section, the totalemission from the source (e.g. in kg s−1) can be deduced.

For around 30 years, the COSPEC instrument has been the principal toolfor remote surveillance of volcanic plumes. Although it was not originallydesigned for volcanological research, it had a major impact on the discipline,not only in volcano monitoring and eruption forecasting but also in estimatingthe global contribution of volcanic volatiles to the atmosphere (e.g. Weibringet al., 1998). By allowing the measurements of SO2 and NO2 flux in airborneplumes, it has fulfilled a vital role in the management of many volcanic crises,notably at Pinatubo in 1991 (Hoff, 1992), and recently at Soufriere Hills Vol-cano. The COSPEC instrument was a major technological innovation at itstime, but is now outdated in several important respects. Problematic aspectsof its concept are interferences from other gases, non-linearity, solar Fraun-hofer lines, polarisation effects, and multiple scattering in aerosol and clouds.In particular, each compound to be measured required its individual mask,and the system is sensitive to wavelength shifts due to mechanical distortionsand temperature variations.

5.7.9 Differential Optical Absorption Spectroscopy (DOAS)

As will be described in detail in the following chapters, the central idea ofDOAS is to make use of structured absorption of many trace gases of atmo-spheric interest while ignoring the rather ‘smooth’ extinction features due to


Fig. 5.3. The principle of a COSPEC instrument (from Barringer et al., 1970)

aerosol or instrumental effects. The DOAS principle and its technical realisa-tion builds on some of the earlier developments, superseding them in manyrespects and supplementing other spectroscopic techniques (e.g. Platt andPerner, 1980; Platt, 1994, 2000; Plane and Smith, 1995; Roscoe and Clemit-shaw, 1997; Russwurm, 1999; Finlayson-Pitts and Pitts, 2000; Clemitshaw,2004).

5 Measurement Techniques for Atmospheric Trace Gas …scarn/teaching/GE4250/PlattStutz... ·...

Documents

Transcript of 5 Measurement Techniques for Atmospheric Trace Gas …scarn/teaching/GE4250/PlattStutz... ·...